updated 1 public sources
large language modelsreasoningreinforcement learninglong-contexttool use

Current frame

Researcher at ByteDance focused on LLM reasoning and reinforcement learning.

Extended note

Kang Liu is a researcher at ByteDance. A public OpenReview profile with a confirmed bytedance.com email lists him as a ByteDance researcher and previously an MS student at Wuhan University from 2014 to 2017. His recent public work includes long-horizon agents, multi-turn RL for LLMs, tool-use reinforcement learning, and reasoning benchmarks.